Paper presented at VSAC 2023

Our Ph.D. fellow Pepe Ballesteros presented his paper “Dialectics of Light” at the 2023 Visual Science of Art Conference

Talk Session: Art with Other Intelligences

Abstract

In words of Leonardo, ‘correctness of light cannot be measured or evaluated by technical means. Its representation is a definite manifestation of artistic talent which, unlike rules, cannot be learned.’ Despite being one of the core elements for rendering space in figurative paintings, it is striking to notice how scant attention pictorial light has received throughout art history literature. Art historians have preferred to emphasize perspective as the major Renaissance achievement rather than light because perspective is more easily defined. Due to light’s abstract nature, and unlike perspective, intention and accident are not easy to distinguish. Description of subtle changes in light features is a challenge not only for language but also for human perception. Visual psychologists have already shown that most of us are not particularly good at judging illumination features in a photograph (nor paintings, by extension).
The present proposal is part of a more significant project which aims to explore the ways in which computational language may help to construct a renewed epistemology to further analyze and name light features in early modern painting. A machine learning model is set up to learn computer graphics-based light features (e.g, Spherical Harmonics) of automatically extracted faces from paintings. Starting from a distant viewing framework, dialectics of light proposes comparing contextualized emerging patterns and established taxonomies of light in art history (e.g, Lomazzo, Wolfgang Schöne). Furthermore, the project envisages the possibility of finding underlying relationships between the visual representation and the description of light encoded in large visual-language models (e.g, CLIP, diffusion-based models). Therefore, dialectics of light strives to highlight the synthesis and contradictions that emerge from the confrontation between machine and human perception.